| Index: trunk/tools/wsor/contribution_inequality/create.sql |
| — | — | @@ -0,0 +1,133 @@ |
| | 2 | +drop table if exists giovanni.policy_contributors; |
| | 3 | + |
| | 4 | +create table giovanni.policy_contributors select |
| | 5 | + page_title as title, |
| | 6 | + page_namespace as namespace, |
| | 7 | + rev_user as user_id, |
| | 8 | + rev_user_text user_name, |
| | 9 | + year(rev_timestamp) as year, |
| | 10 | + count(*) as editcount |
| | 11 | +from revision join page |
| | 12 | +on rev_page = page_id |
| | 13 | +where page_title in ( |
| | 14 | + "Notability", |
| | 15 | + "Verifiability", |
| | 16 | + "No_original_research", |
| | 17 | + "Neutral_point_of_view", |
| | 18 | + "Article_titles", |
| | 19 | + "What_Wikipedia_is_not", |
| | 20 | + "Wikipedia_is_not_a_dictionary", |
| | 21 | + "Biographies_of_living_persons", |
| | 22 | + "Consensus", |
| | 23 | + "Ignore_all_rules", |
| | 24 | + "Deletion_policy", |
| | 25 | + "Criteria_for_speedy_deletion", |
| | 26 | + "Proposed_deletion", |
| | 27 | + "Editing_policy", |
| | 28 | + "Citing_sources", |
| | 29 | + "Disambiguation", |
| | 30 | + "Edit_warring", |
| | 31 | + "Civility", |
| | 32 | + "No_legal_threats", |
| | 33 | + "No_personal_attacks", |
| | 34 | + "Ownership_of_articles", |
| | 35 | + "Sock_puppetry", |
| | 36 | + "Conflict_of_interest", |
| | 37 | + "Do_not_disrupt_Wikipedia_to_illustrate_a_point", |
| | 38 | + "Etiquette", |
| | 39 | + "Gaming_the_system", |
| | 40 | + "Manual_of_Style", |
| | 41 | + "Manual_of_Style_(abbreviations)", |
| | 42 | + "Manual_of_Style_(accessibility)", |
| | 43 | + "Manual_of_Style_(article_message_boxes)", |
| | 44 | + "Manual_of_Style_(biographies)", |
| | 45 | + "Manual_of_Style_(capital_letters)", |
| | 46 | + "Manual_of_Style_(captions)", |
| | 47 | + "Manual_of_Style_(dates_and_numbers)", |
| | 48 | + "Manual_of_Style_(disambiguation_pages)", |
| | 49 | + "Manual_of_Style_(embedded_lists)", |
| | 50 | + "Manual_of_Style_(footnotes)", |
| | 51 | + "Manual_of_Style_(icons)", |
| | 52 | + "Manual_of_Style_(infoboxes)", |
| | 53 | + "Manual_of_Style_(layout)", |
| | 54 | + "Manual_of_Style_(lead_section)", |
| | 55 | + "Manual_of_Style_(linking)", |
| | 56 | + "Manual_of_Style_(lists)", |
| | 57 | + "Manual_of_Style_(lists_of_works)", |
| | 58 | + "Manual_of_Style_(pronunciation)", |
| | 59 | + "Manual_of_Style_(proper_names)", |
| | 60 | + "Manual_of_Style_(self-references_to_avoid)", |
| | 61 | + "Manual_of_Style_(spelling)", |
| | 62 | + "Manual_of_Style_(summary_style)", |
| | 63 | + "Manual_of_Style_(tables)", |
| | 64 | + "Manual_of_Style_(text_formatting)", |
| | 65 | + "Manual_of_Style_(titles)", |
| | 66 | + "Manual_of_Style_(trademarks)", |
| | 67 | + "Manual_of_Style_(trivia_sections)", |
| | 68 | + "Manual_of_Style_(words_to_watch)", |
| | 69 | + "Help_desk", |
| | 70 | + "Reference_desk/Computing", |
| | 71 | + "Reference_desk/Entertainment", |
| | 72 | + "Reference_desk/Humanities", |
| | 73 | + "Reference_desk/Language", |
| | 74 | + "Reference_desk/Mathematics", |
| | 75 | + "Reference_desk/Science", |
| | 76 | + "Reference_desk/Miscellaneous", |
| | 77 | + "Editor%27s_index_to_Wikipedia", |
| | 78 | + "Copyright_problems", |
| | 79 | + "Contributor_copyright_investigations", |
| | 80 | + "External_links/Noticeboard", |
| | 81 | + "Fringe_theories/Noticeboard", |
| | 82 | + "Neutral_point_of_view/Noticeboard", |
| | 83 | + "Reliable_sources/Noticeboard", |
| | 84 | + "Administrator_intervention_against_vandalism", |
| | 85 | + "Arbitration_Committee/Noticeboard", |
| | 86 | + "Arbitration/Requests/Enforcement", |
| | 87 | + "Arbitration/Requests", |
| | 88 | + "Editor_review", |
| | 89 | + "Dispute_resolution_noticeboard", |
| | 90 | + "Wikiquette_assistance", |
| | 91 | + "Requests_for_mediation", |
| | 92 | + "Notability_(academics)", |
| | 93 | + "Notability_(books)", |
| | 94 | + "Notability_(events)", |
| | 95 | + "Notability_(films)", |
| | 96 | + "Notability_(music)", |
| | 97 | + "Notability_(numbers)", |
| | 98 | + "Notability_(people)", |
| | 99 | + "Notability_(sports)", |
| | 100 | + "Notability_(web)", |
| | 101 | + "Notability_(organizations_and_companies)", |
| | 102 | + "No_original_research/Noticeboard", |
| | 103 | + "Fiction/Noticeboard", |
| | 104 | + "Content_forking", |
| | 105 | + "Autobiography", |
| | 106 | + "Non-free_content_review", |
| | 107 | + "Bureaucrats%27_noticeboard", |
| | 108 | + "External_links", |
| | 109 | + "Fringe_theories", |
| | 110 | + "No_disclaimers_in_articles", |
| | 111 | + "Offensive_material", |
| | 112 | + "Patent_nonsense", |
| | 113 | + "Plagiarism", |
| | 114 | + "Spam", |
| | 115 | + "Wikipedia_is_not_for_things_made_up_one_day", |
| | 116 | + "Sexual_content", |
| | 117 | + "Vandalism", |
| | 118 | + "Please_do_not_bite_the_newcomers", |
| | 119 | + "Arbitration/Policy", |
| | 120 | + "Banning_policy", |
| | 121 | + "Blocking_policy", |
| | 122 | + "Administrators", |
| | 123 | + "Bot_policy", |
| | 124 | + "Copyright_violations", |
| | 125 | + "Image_use_policy", |
| | 126 | + "Non-free_content_criteria", |
| | 127 | + "Be_bold", |
| | 128 | + "Categorization/Ethnicity,_gender,_religion_and_sexuality", |
| | 129 | + "Categorization_of_people", |
| | 130 | + "Categorization", |
| | 131 | + "Citing_sources" |
| | 132 | +) |
| | 133 | +and page_namespace in (4,5) |
| | 134 | +group by rev_user_text, year(rev_timestamp), title, namespace |
| Index: trunk/tools/wsor/contribution_inequality/queryall.sh |
| — | — | @@ -0,0 +1,8 @@ |
| | 2 | +#!/bin/bash |
| | 3 | + |
| | 4 | +wd=`pwd` |
| | 5 | +mkdir results |
| | 6 | +pushd results |
| | 7 | +for args in {2001..2010}' '{4..5}; do |
| | 8 | + xargs -I {} $wd/query.py {} $args < $wd/pages.txt |
| | 9 | +done |
| Property changes on: trunk/tools/wsor/contribution_inequality/queryall.sh |
| ___________________________________________________________________ |
| Added: svn:executable |
| 1 | 10 | + * |
| Index: trunk/tools/wsor/contribution_inequality/query.py |
| — | — | @@ -0,0 +1,28 @@ |
| | 2 | +#!/usr/bin/python |
| | 3 | +from argparse import ArgumentParser |
| | 4 | +import os |
| | 5 | +from oursql import connect |
| | 6 | + |
| | 7 | +parser = ArgumentParser() |
| | 8 | +parser.add_argument('page') |
| | 9 | +parser.add_argument('year') |
| | 10 | +parser.add_argument('ns') |
| | 11 | + |
| | 12 | +query=""" |
| | 13 | +select |
| | 14 | + user_name, editcount |
| | 15 | +from giovanni.policy_contributors |
| | 16 | +where title = ? and year = ? and namespace = ? |
| | 17 | +order by editcount desc limit 100; |
| | 18 | +""" |
| | 19 | + |
| | 20 | +if __name__ == '__main__': |
| | 21 | + ns = parser.parse_args() |
| | 22 | + db = connect(read_default_file=os.path.expanduser('~/.my.cnf')) |
| | 23 | + cursor = db.cursor() |
| | 24 | + cursor.execute(query, (ns.page, ns.year, ns.ns)) |
| | 25 | + f = open('%s-%s-%s.tsv' % (ns.page, ns.year, ns.ns), 'w') |
| | 26 | + for row in cursor: |
| | 27 | + print >> f, '%s\t%d' % row |
| | 28 | + print 'results printed to %s' % f.name |
| | 29 | + |
| Property changes on: trunk/tools/wsor/contribution_inequality/query.py |
| ___________________________________________________________________ |
| Added: svn:executable |
| 1 | 30 | + * |